Fake Scientific Papers

Early in March 2014 various news outlets, including Science Recorder and Scientific American reported that over 120 bogus papers had been published by Springer and IEEE. The majority of them concerned computer science and related fields, which explains why IEEE had so many (100 or so).

These fake papers, which were actually nonsensical gibberish, were produced by SCIgen software, which was created by Jeremy Stribling of MIT in about 2005. It uses a data structure of words, phrases, and sentence rules to generate random, grammatically correct, and nonsense sentences. SCIgen, which runs under linux, is available for download, so anyone can use it. There is also an on-line version you can use.

A researcher in France, one Cyril Labbe at Joseph Fourier University in Grenoble, developed a program which can detect the bogus papers. The SCIgen products contain common characteristics of vocabulary and phrasing which are detectable. Labbe found the bogus papers and notified Springer and IEEE. Both entities promptly removed the papers from their systems.

The interesting part of this is that the bogus papers got past editors and reviewers. This is either evidence of the cleverness of SCIgen, weakness in editorial and review mechanisms, or both.